Sharon AI adds confidential computing to sovereign cloud platform

Australian AI infrastructure provider Sharon AI has integrated VAST Data's DataEnclave confidential computing capability into its AI Factory platform, aiming to let regulated organizations run advanced AI models onshore without exposing sensitive data or proprietary model weights to infrastructure operators.

The collaboration, announced September 23, addresses a persistent barrier for banks, government agencies, and other regulated entities across Australia and Asia-Pacific: the inability to use frontier AI models because sensitive data cannot leave controlled environments and model builders won't distribute proprietary models into infrastructure they don't control.

How the technology works

VAST DataEnclave establishes hardware-isolated secure runtimes directly within the VAST AI Operating System. Proprietary models execute inside confidential virtual machines built on CPU-level trusted execution environments, with NVIDIA GPUs operating in Confidential Compute mode. Guest memory, GPU memory, and interconnect traffic remain encrypted throughout processing.

The system uses cryptographic attestation to verify the runtime environment and its enforced policy before sensitive assets are decrypted and loaded into a secure enclave. Sharon AI and other infrastructure operators cannot access customer data or model weights while they're being processed.

Customers and model builders retain independent control over their respective encryption keys through Bring Your Own KMS integrations. Customer data keys stay within the customer's trust domain, and model keys remain within the model builder's trust domain at all times. The platform can run connected through DataEnclave's attestation service or fully air-gapped with on-premises attestation and key brokering for organizations that cannot connect to external services.

Every attestation event, key release, and enclave lifecycle action is recorded in a tamper-proof audit trail, giving customers visibility into what ran, where, and under what verified policy.

Why it matters

Confidential computing addresses a fundamental tension in enterprise AI adoption: organizations in regulated industries need access to the most capable models, but those models are typically only available through cloud services that require sending data offshore or to third-party infrastructure. This creates compliance risks and limits AI deployment in sectors like financial services, healthcare, and government.

By combining hardware-level isolation with cryptographic attestation, the technology enables a new model where neither the infrastructure operator nor the model provider can access the other party's intellectual property during processing. This should expand the range of AI workloads regulated organizations can run while maintaining data sovereignty requirements.

"Our customers across Australia and Asia-Pacific need to run AI at full speed without compromising on data sovereignty, and increasingly they also want access to frontier models that were previously only available offshore," said James Manning, co-founder and CEO of Sharon AI. "Building on our sovereign data foundation with VAST DataEnclave is expected to let us host those models onshore, inside attested environments where the model owner's weights and the customer's data are both protected from everyone, including us."

Sunil Chavan, vice president for Asia-Pacific and Japan at VAST Data, noted that the collaboration represents "exactly the kind of regional AI infrastructure where VAST DataEnclave may be able to change what's possible, giving customers across Australia and Asia-Pacific access to leading AI models while allowing both sides to retain control of their most valuable intellectual property."

The integration extends Sharon AI's sovereign AI Factory platform to support attested, confidential AI workloads alongside its existing security, governance, and compliance capabilities. Details were first reported by Sharon AI in a press release.